Measures of Central Tendency: Mean, Median, Mode

Not sure you’re ready?

Take the ~3-minute readiness diagnostic and see where you stand.

Imagine trying to summarize the mass of an irregularly shaped asteroid drifting through space. Instead of tracking every jagged rock and crater, a physicist pinpoints a single mathematical coordinate—the center of mass—to predict how the entire object will move. In statistics, a measure of central tendency serves the exact same purpose for a set of data. It provides a single summary value representing the center point of a data distribution, cutting through the noise of individual variations to capture the essence of the group.

Just as the physical center of mass allows this toy bird to balance on a single point, a measure of central tendency provides a single balancing point for a statistical distribution.

However, "the center" is not a singular, fixed concept. Depending on the shape of the data and the anomalies hidden within it, we must choose our tools carefully. We rely on three distinct mathematical centers—the mean, the median, and the mode—each of which responds differently to the forces within a dataset. Understanding not just how to calculate these measures, but how they behave under mathematical transformations and graphical distortions, is fundamental to statistical literacy.

When most people use the word "average" in daily life, they are referring to the arithmetic mean. It is the mathematical center of gravity of a dataset.

Definition: The arithmetic mean is calculated by finding the sum of all values in a data set and dividing the sum by the total number of values.

Because it incorporates the exact magnitude of every single data point, the arithmetic mean is highly precise. In statistical notation, we use specific symbols to denote whether we are talking about an entire population or just a subset. The formula for a population mean is represented in statistics by the Greek letter mu ( $\mu$ ). Conversely, when dealing with a sample, the formula for a sample mean is represented in statistics by the letter x with a horizontal bar over the letter ( $\bar{x}$ ).

Despite its precision, the arithmetic mean has a profound physical vulnerability: it behaves like a lever. The arithmetic mean is sensitive to extremely high or extremely low values in a data set. We call these extremely high or extremely low values outliers.

Because the mean mathematically balances the distances between all data points, an outlier acts like a heavy weight placed at the far end of a seesaw. By necessity, an outlier pulls the arithmetic mean toward the outlier's numerical value. If an exam is given to a class of ten students, and nine students score an 85 while one student scores a 10, that single extremely low value will drag the arithmetic mean down to 77.5, misrepresenting the performance of the vast majority of the class.

The arithmetic mean balances a dataset exactly like a physical lever. A heavy outlier placed far from the center will drastically pull the mean toward it to maintain mathematical balance.

To solve the vulnerability of the mean, we turn to a measure based on position rather than magnitude.

Definition: The median represents the exact middle numerical value of a given data set.

The median does not care how large or small the numbers at the edges are; it only cares about the physical center of the line. However, to find the median, the values in a data set must first be arranged in ascending or descending numerical order. If you do not sort the data first, you are merely picking a random number from an unorganized pile.

Once the data is ordered, the calculation branches into two simple rules based on the size of the dataset:

Odd Number of Values: For a data set with an odd number of values, the median is the single middle value in the ordered list. (e.g., In the set {2, 4, 7, 9, 12}, the median is 7).
Even Number of Values: For a data set with an even number of values, the exact center falls between two numbers. Therefore, the median is the arithmetic mean of the two middle values in the ordered list. (e.g., In the set {2, 4, 6, 8, 10, 12}, the median is $\frac{6 + 8}{2} = 7$ ).

Calculating the median requires first sorting the data. In an odd dataset, it is the exact middle value; in an even dataset, it is the arithmetic mean of the two middle values.

Because the median relies purely on rank order, the median is highly resistant to extreme values, or outliers, in a data set. If we take the dataset {1, 2, 3, 4, 5}, the median is 3. If we change the 5 to a 5,000, creating the set {1, 2, 3, 4, 5000}, the median remains exactly 3. The center position has not changed, making the median the superior metric when evaluating highly skewed data, such as household incomes or home prices.

The third measure of central tendency asks a completely different question: What happens most often?

Definition: The mode is the value or values occurring most frequently in a data set.

Unlike the mean and median, which always yield a single valid answer, the mode is entirely dependent on repetition. This leads to three distinct scenarios:

No Mode: A data set has no mode if every value in the data set appears with the exact same frequency. If your data set is {10, 12, 14, 16}, no single number dominates.
Bimodal: A data set is considered bimodal if exactly two distinct values share the highest frequency of occurrence. (e.g., In {2, 2, 5, 8, 8}, the modes are 2 and 8).
Multimodal: A data set is considered multimodal if more than two distinct values share the highest frequency of occurrence.

A bimodal data distribution features two distinct peaks, indicating two different values that share the highest frequency of occurrence.

The mode possesses one extraordinary superpower that the mean and median fundamentally lack. The mode is the only measure of central tendency applicable to categorical, non-numerical data. If a botanist is trying to find the "center" of a dataset consisting of flower colors—{Red, Blue, Blue, Yellow, Red, Blue}—you cannot calculate the arithmetic mean of a color, nor can you logically rank them to find a median. You can, however, state that "Blue" is the mode.

When we graph a large dataset—perhaps plotting the heights of a thousand trees—the shape of the resulting curve tells us immediately how the mean, median, and mode will interact.

If the forces of nature act uniformly, we often see a classic bell curve. In a perfectly symmetrical, unimodal data distribution, the arithmetic mean, median, and mode share the exact same numerical value. They sit perfectly atop one another at the exact center of the bell.

In a perfectly symmetrical normal distribution, the arithmetic mean, median, and mode are identical and perfectly aligned at the center.

However, when data is asymmetrical, the distribution is "skewed," and our three measures pull apart.

Right-Skewed Distributions

Imagine a small town where the average income is $50,000, but a billionaire moves in. That massive outlier sits far to the right on a number line, stretching the "tail" of the graph to the right.

Because the outlier pulls the mean toward its value, in a right-skewed data distribution, the arithmetic mean is typically greater than the median. The median stays anchored in the middle, while the mean chases the billionaire.

Left-Skewed Distributions

Conversely, imagine a college exam where most students score between 80 and 100, but a few students completely sleep through the test and score zeros. The tail of the graph stretches to the left.

In a left-skewed data distribution, the arithmetic mean is typically less than the median. The heavy anchor of the zeros pulls the arithmetic mean down, while the median remains safely nestled among the passing grades.

In skewed distributions, the long tail of outliers pulls the arithmetic mean further away from the peak (mode) than the median, separating the three measures of central tendency.

Distribution Shape	Visual Characteristic	Relationship of Measures
Perfectly Symmetrical	Bell-shaped, perfectly balanced	Mean = Median = Mode
Right-Skewed	Long tail extends to the right (high values)	Mean > Median
Left-Skewed	Long tail extends to the left (low values)	Mean < Median

One of the most beautiful properties of central tendency is how predictably it behaves when we alter the underlying data. In statistics, we frequently perform linear transformations—such as adding a curve to exam scores or converting measurements from inches to centimeters.

How do these operations affect the center? The rule is beautifully absolute: whatever linear operation you apply to every data point, you apply exactly the same operation to the mean, median, and mode.

Shifting the Data (Addition and Subtraction)

If a teacher decides a test was too difficult and adds 5 points to every student's score, the entire shape of the data simply slides 5 points to the right on the number line. The relative distance between students does not change.

Adding a constant value to every number in a data set increases the arithmetic mean by that exact constant value.
Adding a constant value to every number in a data set increases the median by that exact constant value.
Adding a constant value to every number in a data set increases the mode by that exact constant value.

Naturally, the reverse is true if we deduct points:

Subtracting a constant value from every number in a data set decreases the arithmetic mean by that exact constant value.
Subtracting a constant value from every number in a data set decreases the median by that exact constant value.
Subtracting a constant value from every number in a data set decreases the mode by that exact constant value.

Scaling the Data (Multiplication and Division)

Scaling data behaves similarly, but instead of sliding the distribution, it stretches or compresses it. For instance, if we convert a dataset of lengths from feet to inches, we multiply every single value by 12.

Multiplying every number in a data set by a constant value multiplies the arithmetic mean by that exact constant value.
Multiplying every number in a data set by a constant value multiplies the median by that exact constant value.
Multiplying every number in a data set by a constant value multiplies the mode by that exact constant value.

And if we compress the data, say by converting from centimeters to meters (dividing by 100):

Dividing every number in a data set by a non-zero constant value divides the arithmetic mean by that exact constant value.
Dividing every number in a data set by a non-zero constant value divides the median by that exact constant value.
Dividing every number in a data set by a non-zero constant value divides the mode by that exact constant value.

Summary of Central Tendency

Mastering measures of central tendency requires looking past the formulas and recognizing the underlying geometry of the data. The mean provides a sensitive, inclusive balance point, susceptible to the gravity of outliers. The median provides a stubborn, unyielding center, perfect for cutting through skewed extremities. The mode tracks the frequency of occurrence, standing alone as the tool for categorical observation. Together, they allow us to compress thousands of chaotic data points into a singular, highly legible narrative of truth.